VAT08 Stage 1 and 2 Data

nAb missingness

Missingness across variants. Some ptids only have the ancestral marker measured. The missingness pattern is explained by the fact that samples from different steps have different set of markers measured. We decide to impute all variants if one variant marker exists.

Overall, “because the correlation among variants is pretty good for the non-naives and really the analysis focuses on the non-naives” (Peter), we decide to impute variants for those who have data for at least one variant to make the complexity of data processing more manageable.

dat1=dat_mapped
# the pattern in cases is un-remarkable
# dat1=subset(dat_mapped, 1==EventIndPrimaryD1)
for (t in 2:1) {
  cat(paste0("stage ", t, '\n'))
  for (tp in c("B", "Day22", "Day43")) {
    # placebo has a similar pattern
    for(trt in 1) {
      myprint(tp, trt)
      dat=subset(dat1, Trialstage==t & ph1.D43 & Trt==trt) # pattern is independent of Bserostatus
      for (i in 1:5) {
        dat[['tmp'%.%i]]=ifelse(is.na(dat[[tp%.%assays[i]]]), 0, 1)
      }
      dat$tmp=with(dat,paste0(tmp1,tmp2,tmp3,tmp4,tmp5))
      print(table(dat$tmp))
      cat("\n")
    }
  }
}

## stage 2
## tp = B ; trt = 1 
## 
## 00000 11111 
##  4619   485 
## 
## tp = Day22 ; trt = 1 
## 
## 00000 11111 
##  4620   484 
## 
## tp = Day43 ; trt = 1 
## 
## 00000 11111 
##  4621   483 
## 
## stage 1
## tp = B ; trt = 1 
## 
## 00000 11111 
##  3605   487 
## 
## tp = Day22 ; trt = 1 
## 
## 00000 01111 10111 11111 
##  3604     1     1   486 
## 
## tp = Day43 ; trt = 1 
## 
## 00000 11111 
##  3608   484

Missingness across baseline, d22 and d43, for the ancestral markers in the vaccine arms. The results support the decision “to keep it simple and require for RIS membership availability of data at all 3 time points” (Peter).

for(trt in 1:0) {
  myprint(trt)
  for (i in 2:1) {
    cat(paste0("stage ", i, '\n'))
    dat=subset(dat_proc, Trialstage==i & Trt==trt & ph1.D43)
    times=c("B","Day22","Day43")
    for (i in 1:3) {
      dat[['tmp'%.%i]]=ifelse(is.na(dat[[times[i]%.%'pseudoneutid50']]), 0, 1)
    }
    dat$tmp=with(dat,paste0(tmp1,tmp2,tmp3))
    print(table(dat$tmp))
    cat("\n")
  }
}

## trt = 1 
## stage 2
## 
##  000  001  011  101  110  111 
## 4338    4    6    1   11  713 
## 
## stage 1
## 
##  000  010  011  100  101  110  111 
## 2961   12   28   45   15   29  956 
## 
## trt = 0 
## stage 2
## 
##  000  001  011  100  101  110  111 
## 4377    2    5    1    3    2  562 
## 
## stage 1
## 
##  000  001  010  011  100  101  110  111 
## 3210    2    3   10   31   10   24  827

# a more detailed look at missingness by batch
# for (tp in c("B")) {
# #for (tp in c("B", "Day22", "Day43")) {
#   # before imputation
#   dat1=dat_mapped
#   # after imputation
#   # dat1=dat_proc
#   
#   for(trt in 1) {
#     myprint(tp, trt)
#     for (i in 2:1) { # stage
#       cat(paste0("stage ", i, '\n'))
#       if (trt==2) {
#         # pool over arms
#         dat=subset(dat1, Trialstage==i & ph1.D43)
#       } else {
#         dat=subset(dat1, Trialstage==i & ph1.D43 & Trt==trt)
#       }
#       for (i in 1:5) {
#         dat[['tmp'%.%i]]=ifelse(is.na(dat[[tp%.%assays[i]]]), 0, 1)
#       }
#       dat$tmp=with(dat,paste0(tmp1,tmp2,tmp3,tmp4,tmp5))
#       print(mytable(dat$EventIndPrimaryD43, dat$tmp, dat$nAbBatch, dat$Bserostatus))
#     }
#   }
# 
# }

The number of D43 nAb ph2 samples by Bserostatus 0/1 (row) and Case 1/0 (column) and broken down by treatment arms (vaccine, placebo).

dat1=dat_proc

f=function(tab2, tab1, tab0) {
  for (i in 1:2) {
    print(paste0(tab2[i,1], " (", tab1[i,1], ", ", tab0[i,1], ") ", tab2[i,2], " (", tab1[i,2], ", ", tab0[i,2], ")"))
  }
}

for (i in 2:1) {
  cat(paste0("stage ", i, '\n'))
  dat=subset(dat1, Trialstage==i & ph1.D43 & TwophasesampIndD43nAb)
  tab2 = table(dat$Bserostatus, dat$EventIndPrimaryD43)[,2:1] # [,2:1] moves cases before non-cases
  dat=subset(dat1, Trialstage==i & ph1.D43 & Trt==1 & TwophasesampIndD43nAb)
  tab1 = table(dat$Bserostatus, dat$EventIndPrimaryD43)[,2:1] # [,2:1] moves cases before non-cases
  dat=subset(dat1, Trialstage==i & ph1.D43 & Trt==0 & TwophasesampIndD43nAb)
  tab0 = table(dat$Bserostatus, dat$EventIndPrimaryD43)[,2:1] # [,2:1] moves cases before non-cases
  f(tab2, tab1, tab0)
}

## stage 2
## [1] "41 (17, 24) 232 (179, 53)"
## [1] "122 (61, 61) 884 (457, 427)"
## stage 1
## [1] "150 (74, 76) 630 (382, 248)"
## [1] "179 (68, 111) 849 (447, 402)"

nAb correlation

Correlation is high between variant markers in Stage 2 non-naive, vaccine arm at both baseline and D43

mypairs(subset(dat_proc, Trialstage==2 & Bserostatus==1 & Trt==1 & ph1.D43)[,"B"%.%nassays])

mypairs(subset(dat_proc, Trialstage==2 & Bserostatus==1 & Trt==1 & ph1.D43)[,"Day43"%.%nassays])

Correlation is medium between variant markers at D43 in Stage 2 naive, vaccine arm. (The correlation between the ancestral marker and the other markers range between 0.63 and 0.67.)

mypairs(subset(dat_proc, Trialstage==2 & Bserostatus==0 & Trt==1 & ph1.D43)[,"Day43"%.%nassays])

Correlation is low between baseline and D22 or D43 in Stage 2 non-naive, vaccine arm.

mypairs(subset(dat_proc, Trialstage==2 & Bserostatus==1 & Trt==1 & ph1.D43)[,c("B","Day22","Day43")%.%assays[1]])

In stage 1, correlation is medium between ancestral markers and other markers. Correlation is high between BA4BA5 and BA1 among non-naive, but medium among naive.

mypairs(subset(dat_proc, Trialstage==1 & Bserostatus==1 & Trt==1 & ph1.D43)[,"Day22"%.%nassays])

mypairs(subset(dat_proc, Trialstage==1 & Bserostatus==0 & Trt==1 & ph1.D43)[,"Day22"%.%nassays])

A closer look at the relationship between baseline and Day 43 markers in the baseline positive. Note that the D43 markers appear saturated since level seems flat across different baseline marker levels.

par(mfrow=c(2,4))
  # ancestral marker
  corplot(Day43pseudoneutid50~Bpseudoneutid50, subset(dat_proc, Trialstage==2 & Bserostatus==1 & Trt==1 & ph1.D43), ylim=c(0,4))
  myboxplot(Day43pseudoneutid50~I(Bpseudoneutid50>0.2), subset(dat_proc, Trialstage==2 & Bserostatus==1 & Trt==1 & ph1.D43), names=c("Undetectable at B","Detectable at B"), ylab="D43", ylim=c(0,4))
  my.interaction.plot(subset(dat_proc, Trialstage==2 & Bserostatus==1 & Trt==1 & ph1.D43 & !is.na(Bpseudoneutid50) & !is.na(Day43pseudoneutid50), c(Bpseudoneutid50, Day43pseudoneutid50)), 
                      x.ori = 0, xaxislabels = c("B", "D43"), cex.axis = 1, add = FALSE, xlab = "", ylab = "", pcol = NULL, lcol = NULL)  
  corplot(Delta43overBpseudoneutid50~Bpseudoneutid50, subset(dat_proc, Trialstage==2 & Bserostatus==1 & Trt==1 & ph1.D43))
  
  # BA.1
  corplot(Day43pseudoneutid50_BA.1~Bpseudoneutid50_BA.1, subset(dat_proc, Trialstage==2 & Bserostatus==1 & Trt==1 & ph1.D43), ylim=c(0,4))
  myboxplot(Day43pseudoneutid50_BA.1~I(Bpseudoneutid50_BA.1>0.2), subset(dat_proc, Trialstage==2 & Bserostatus==1 & Trt==1 & ph1.D43), names=c("Undetectable at B","Detectable at B"), ylab="D43", ylim=c(0,4))
  my.interaction.plot(subset(dat_proc, Trialstage==2 & Bserostatus==1 & Trt==1 & ph1.D43 & !is.na(Bpseudoneutid50_BA.1) & !is.na(Day43pseudoneutid50_BA.1), c(Bpseudoneutid50_BA.1, Day43pseudoneutid50_BA.1)), 
                      x.ori = 0, xaxislabels = c("B", "D43"), cex.axis = 1, add = FALSE, xlab = "", ylab = "", pcol = NULL, lcol = NULL)  
  corplot(Delta43overBpseudoneutid50_BA.1~Bpseudoneutid50_BA.1, subset(dat_proc, Trialstage==2 & Bserostatus==1 & Trt==1 & ph1.D43))

bAb missingness

Missingness across variants is mostly all or none. We decide to impute all variants if one variant marker exists.

# before imputation
dat1=dat_mapped
# after imputation
# dat1=dat_proc
for (t in 2:1) {
  cat(paste0("stage ", t, '\n'))
  # for (tp in c("B")) {
  for (tp in c("B", "Day22", "Day43")) {
    
    # placebo has a similar patten
    for(trt in 1) {
      myprint(tp, trt)
      if (trt==2) {
        # pool over arms
        dat=subset(dat1, Trialstage==t & ph1.D43)
      } else {
        dat=subset(dat1, Trialstage==t & ph1.D43 & Trt==trt)
      }
      for (i in 1:8) {
        dat[['tmp'%.%i]]=ifelse(is.na(dat[[tp%.%assays[i]]]), 0, 1)
      }
      dat$tmp=with(dat,paste0(tmp1,tmp2,tmp3,tmp4,tmp5,tmp6,tmp7,tmp8))
      print(table(dat$tmp))
      cat("\n")
      # print(table(dat$EventIndPrimaryD43, dat$tmp, dat$Bserostatus))
    }
  }

}

## stage 2
## tp = B ; trt = 1 
## 
## 00000000 11111111 
##     4619      485 
## 
## tp = Day22 ; trt = 1 
## 
## 00000000 11111111 
##     4620      484 
## 
## tp = Day43 ; trt = 1 
## 
## 00000000 11111111 
##     4621      483 
## 
## stage 1
## tp = B ; trt = 1 
## 
## 00000000 11111111 
##     3605      487 
## 
## tp = Day22 ; trt = 1 
## 
## 00000000 01111110 10111111 11111101 11111110 11111111 
##     3604        1        1        1        5      480 
## 
## tp = Day43 ; trt = 1 
## 
## 00000000 11111111 
##     3608      484

Missingness across baseline, d22 and d43, for the ancestral markers in the vaccine arms.

for(trt in 1:0) {
  myprint(trt)
  for (i in 2:1) {
    cat(paste0("stage ", i, '\n'))
    dat=subset(dat_proc, Trialstage==i & Trt==trt & ph1.D43)
    times=c("B","Day22","Day43")
    for (i in 1:3) {
      dat[['tmp'%.%i]]=ifelse(is.na(dat[[times[i]%.%'bindSpike']]), 0, 1)
    }
    dat$tmp=with(dat,paste0(tmp1,tmp2,tmp3))
    print(table(dat$tmp))
    cat("\n")
  }
}

## trt = 1 
## stage 2
## 
##  000  101  110  111 
## 4590    1    2  480 
## 
## stage 1
## 
##  000  010  011  100  101  110  111 
## 3560    2    1    1    1    3  478 
## 
## trt = 0 
## stage 2
## 
##  000  111 
## 4575  377 
## 
## stage 1
## 
##  000  011  101  110  111 
## 3697    3    2   11  404

bAb correlation

Correlation is high between variant markers in Stage 2 non-naive, vaccine arm at baseline

mypairs(subset(dat_proc, Trialstage==2 & Bserostatus==1 & Trt==1 & ph1.D43)[,"B"%.%bassays])

and D43

mypairs(subset(dat_proc, Trialstage==2 & Bserostatus==1 & Trt==1 & ph1.D43)[,"Day43"%.%bassays])

Correlation is still high between variant markers at D43 in Stage 2 naive, vaccine arm. (The correlation between the ancestral marker and the other markers range between 0.63 and 0.67.)

mypairs(subset(dat_proc, Trialstage==2 & Bserostatus==0 & Trt==1 & ph1.D43)[,"Day43"%.%bassays])

Correlation is low between baseline and D22 or D43 in Stage 2 non-naive, vaccine arm.

mypairs(subset(dat_proc, Trialstage==2 & Bserostatus==1 & Trt==1 & ph1.D43)[,c("B","Day22","Day43")%.%assays[1]])

mypairs(subset(dat_proc, Trialstage==2 & Bserostatus==1 & Trt==1 & ph1.D43)[,c("B","Day22","Day43")%.%assays[8]])

A closer look at the relationship between baseline and Day 43 markers in the baseline positive. Note that the D43 markers appear saturated since level seems flat across different baseline marker levels.

par(mfrow=c(2,4))
  # ancestral marker
  corplot(Day43bindSpike~BbindSpike, subset(dat_proc, Trialstage==2 & Bserostatus==1 & Trt==1 & ph1.D43), ylim=c(0,4))
  myboxplot(Day43bindSpike~I(BbindSpike>0.5), subset(dat_proc, Trialstage==2 & Bserostatus==1 & Trt==1 & ph1.D43), names=c("Undetectable at B","Detectable at B"), ylab="D43", ylim=c(0,4))
  my.interaction.plot(subset(dat_proc, Trialstage==2 & Bserostatus==1 & Trt==1 & ph1.D43 & !is.na(BbindSpike) & !is.na(Day43bindSpike), c(BbindSpike, Day43bindSpike)), 
                      x.ori = 0, xaxislabels = c("B", "D43"), cex.axis = 1, add = FALSE, xlab = "", ylab = "", pcol = NULL, lcol = NULL)  
  corplot(Delta43overBbindSpike~BbindSpike, subset(dat_proc, Trialstage==2 & Bserostatus==1 & Trt==1 & ph1.D43))
  
  # omicron
  corplot(Day43bindSpike_omicron~BbindSpike_omicron, subset(dat_proc, Trialstage==2 & Bserostatus==1 & Trt==1 & ph1.D43), ylim=c(0,4))
  myboxplot(Day43bindSpike_omicron~I(BbindSpike_omicron>0.5), subset(dat_proc, Trialstage==2 & Bserostatus==1 & Trt==1 & ph1.D43), names=c("Undetectable at B","Detectable at B"), ylab="D43", ylim=c(0,4))
  my.interaction.plot(subset(dat_proc, Trialstage==2 & Bserostatus==1 & Trt==1 & ph1.D43 & !is.na(BbindSpike_omicron) & !is.na(Day43bindSpike_omicron), c(BbindSpike_omicron, Day43bindSpike_omicron)), 
                      x.ori = 0, xaxislabels = c("B", "D43"), cex.axis = 1, add = FALSE, xlab = "", ylab = "", pcol = NULL, lcol = NULL)  
  corplot(Delta43overBbindSpike_omicron~BbindSpike_omicron, subset(dat_proc, Trialstage==2 & Bserostatus==1 & Trt==1 & ph1.D43))

Misc

missingness pattern across nAb and bAb markers shows that there is no overlap in pattern between nAb and nAb in stage 1 or 2.

dat1=dat_mapped
for (t in 2:1) {
  cat(paste0("stage ", t, '\n'))
  for (tp in c("Day43")) {

    # placebo has a similar pattern
    for(trt in 1) {
      myprint(tp, trt)
      dat=subset(dat1, Trialstage==t & ph1.D43 & Trt==trt)
      
      dat$tmp1 = ifelse(is.na(dat$Day43pseudoneutid50), 0, 1)
      dat$tmp2 = ifelse(is.na(dat$Day43pseudoneutid50_B.1.351), 0, 1)
      dat$tmp3 = ifelse(is.na(dat$Day43bindSpike), 0, 1)
      
      dat$tmp=with(dat,paste0(tmp1,tmp2))
      print(table(dat$tmp))
      dat$tmp=with(dat,paste0(tmp1,tmp3))
      print(table(dat$tmp))
      dat$tmp=with(dat,paste0(tmp2,tmp3))
      print(table(dat$tmp))
      cat("\n")
    }
  }
}

## stage 2
## tp = Day43 ; trt = 1 
## 
##   00   01   10   11 
## 4382    3   20  699 
## 
##   00   01   10   11 
## 4365   20  256  463 
## 
##   00   01   10   11 
## 4369   33  252  450 
## 
## stage 1
## tp = Day43 ; trt = 1 
## 
##   00   01   10   11 
## 3074    5  431  582 
## 
##   00   01   10   11 
## 3072    7  536  477 
## 
##   00   01   10   11 
## 3472   33  136  451

Omicron bAb ULOQ may be too low.

corplot(Day43bindSpike_omicron~Day43bindSpike, subset(dat_proc, Trialstage==2 & Bserostatus==1 & Trt==1 & ph1.D43), add.diagonal.line = F, method="pearson")

corplot(Day43pseudoneutid50_BA.4.5~Day43pseudoneutid50, subset(dat_proc, Trialstage==2 & Bserostatus==1 & Trt==1 & ph1.D43), add.diagonal.line = F, method="pearson")

correlation and missingness across bAb and nAb

  for (sero in c(0,1)) {    
    for (trt in c(0,1)) {
      for (stage in c(1,2)) {
        myprint(stage, trt, sero)
        dat=subset(dat_mapped, Trialstage==stage & Trt==trt & Bserostatus==sero)
        corplot(Day43bindSpike~Day43pseudoneutid50, dat)
        print(table(!is.na(dat$Day43bindSpike), !is.na(dat$Day43pseudoneutid50), dat$EventIndPrimaryD43))
      }
    }
}

## stage = 1 ; trt = 0 ; sero = 0

## , ,  = 0
## 
##        
##         FALSE TRUE
##   FALSE   555  246
##   TRUE      0   37
## 
## , ,  = 1
## 
##        
##         FALSE TRUE
##   FALSE     7   10
##   TRUE      0   85
## 
## stage = 2 ; trt = 0 ; sero = 0

## , ,  = 0
## 
##        
##         FALSE TRUE
##   FALSE   263   36
##   TRUE      2   32
## 
## , ,  = 1
## 
##        
##         FALSE TRUE
##   FALSE     1    1
##   TRUE      3   35
## 
## stage = 1 ; trt = 1 ; sero = 0

## , ,  = 0
## 
##        
##         FALSE TRUE
##   FALSE   379  271
##   TRUE      4  239
## 
## , ,  = 1
## 
##        
##         FALSE TRUE
##   FALSE     4    6
##   TRUE      0   93
## 
## stage = 2 ; trt = 1 ; sero = 0

## , ,  = 0
## 
##        
##         FALSE TRUE
##   FALSE    93   55
##   TRUE     14  166
## 
## , ,  = 1
## 
##        
##         FALSE TRUE
##   FALSE     0    0
##   TRUE      5   23
## 
## stage = 1 ; trt = 0 ; sero = 1

## , ,  = 0
## 
##        
##         FALSE TRUE
##   FALSE  2785  210
##   TRUE     21  183
## 
## , ,  = 1
## 
##        
##         FALSE TRUE
##   FALSE     2    7
##   TRUE      9   96
## 
## stage = 2 ; trt = 0 ; sero = 1

## , ,  = 0
## 
##        
##         FALSE TRUE
##   FALSE  4170  166
##   TRUE     14  244
## 
## , ,  = 1
## 
##        
##         FALSE TRUE
##   FALSE     6    2
##   TRUE      9   59
## 
## stage = 1 ; trt = 1 ; sero = 1

## , ,  = 0
## 
##        
##         FALSE TRUE
##   FALSE  2741  306
##   TRUE      4  166
## 
## , ,  = 1
## 
##        
##         FALSE TRUE
##   FALSE     3    6
##   TRUE      0   63
## 
## stage = 2 ; trt = 1 ; sero = 1

## , ,  = 0
## 
##        
##         FALSE TRUE
##   FALSE  4265  209
##   TRUE      9  248
## 
## , ,  = 1
## 
##        
##         FALSE TRUE
##   FALSE     5    1
##   TRUE      1   64

Note that the marker level in the NN, placebo is higher in Stage 2 than in Stage 1, probably reflecting different countries.

myboxplot(Day43bindSpike~Trt+Bserostatus+Trialstage, dat_mapped, ylab="D43 bAb Spike",
  names=rep(c("N,pla", "N,vac", "NN,pla", "NN,vac"),2),
  main="Stage 1                                        Stage 2",
  cex.axis=0.8)
abline(v=4.5)

Check which ptids are in which batch

We have list of ptids sampled in step 2 in both stages of the trial.

stage2step2=read.csv("~/Aiying700.csv")
dat_mapped$stage2step2 = dat_mapped$Subjectid %in% stage2step2$Unique.Subject.Identifier
dat_proc$stage2step2 = dat_proc$Ptid %in% stage2step2$Unique.Subject.Identifier

stage1step2=read.csv("~/JinStage1Step2.csv")
dat_mapped$stage1step2 = dat_mapped$SUBJID %in% stage1step2$SUBJID
dat_proc$stage1step2 = dat_proc$SUBJID %in% stage1step2$SUBJID

Stage 2, nAb. batch 1 = step2 for both cases and noncases

with(subset(dat_proc, Trt==1 & Trialstage==2 & ph2.D43.nAb), table(stage2step2, EventIndPrimaryD1, nAbBatch))

## , , nAbBatch = 1
## 
##            EventIndPrimaryD1
## stage2step2   0   1
##       FALSE   0   0
##       TRUE  286  24
## 
## , , nAbBatch = 2
## 
##            EventIndPrimaryD1
## stage2step2   0   1
##       FALSE 350  54
##       TRUE    0   0

with(subset(dat_proc, Trt==0 & Trialstage==2 & ph2.D43.nAb), table(stage2step2, EventIndPrimaryD1, nAbBatch))

## , , nAbBatch = 1
## 
##            EventIndPrimaryD1
## stage2step2   0   1
##       FALSE   0   0
##       TRUE  229  30
## 
## , , nAbBatch = 2
## 
##            EventIndPrimaryD1
## stage2step2   0   1
##       FALSE 251  55
##       TRUE    0   0

with(subset(dat_proc, Trt==1 & Trialstage==2 & ph2.D22.nAb), table(stage2step2, EventIndPrimaryD1, nAbBatch))

## , , nAbBatch = 1
## 
##            EventIndPrimaryD1
## stage2step2   0   1
##       FALSE   0   0
##       TRUE  286  40
## 
## , , nAbBatch = 2
## 
##            EventIndPrimaryD1
## stage2step2   0   1
##       FALSE 358  58
##       TRUE    0   0

with(subset(dat_proc, Trt==0 & Trialstage==2 & ph2.D22.nAb), table(stage2step2, EventIndPrimaryD1, nAbBatch))

## , , nAbBatch = 1
## 
##            EventIndPrimaryD1
## stage2step2   0   1
##       FALSE   0   0
##       TRUE  228  81
## 
## , , nAbBatch = 2
## 
##            EventIndPrimaryD1
## stage2step2   0   1
##       FALSE 254  57
##       TRUE    0   0

Stage 1, nAb. Most non-cases are done in batch 2, most cases are done in batch 1. Step 1 and 2 are spread mostly equally between batches.

with(subset(dat_proc, Trialstage==1 & ph2.D43.nAb), table(stage1step2, nAbBatch, EventIndPrimaryD1))

## , , EventIndPrimaryD1 = 0
## 
##            nAbBatch
## stage1step2   1   2
##       FALSE  24 576
##       TRUE    5 175
## 
## , , EventIndPrimaryD1 = 1
## 
##            nAbBatch
## stage1step2   1   2
##       FALSE 221  68
##       TRUE   32   6

bAb.

# stage 1
with(subset(dat_proc, Trialstage==1 & ph2.D22.bAb), table(stage1step2, EventIndPrimaryD1))

##            EventIndPrimaryD1
## stage1step2   0   1
##       FALSE 478 298
##       TRUE  129  43

# stage 2
with(subset(dat_proc, Trialstage==2 & ph2.D22.bAb), table(stage2step2, EventIndPrimaryD1))

##            EventIndPrimaryD1
## stage2step2   0   1
##       FALSE 341 126
##       TRUE  347 125